|
|
|
|
|
|
|
Traditionally, the end of a century has been a time of doomsday predictions and unrest - both imagined and real. According to Dutch bank ING Barings, the British and Dutch stock markets fell sharply in 1699 and 1799. At the end of 1899, the Dow fell 23 percent, and prices continued to drop sharply for several weeks into New Year 1900. As one writer for The Economist observed, "A millennium, even more than a centennial, would be spooky enough without the fear of a computer failure."
Media hype about Y2K, much of which strains credibility, has become common. It is not unusual to hear expressions of surprise or, unfortunately, boredom from people who should be paying attention. "You're really serious about this!" was a comment heard by Year 2000 Team members who recently staffed a Lobby 10 Y2K exhibit. MIT is serious about potential Year 2000 computer and microchip failures. Over the coming months, this column will address several topics related to MIT's Y2K preparations.
The basis for computer failure is simple to describe, though the effects are not. Computer hardware, software, and microchips embedded in electronic devices may interpret the year 2000 as 1900 – or as a date that is undefined – because they store the year in two-digit form. In addition, the year 2000 (unlike most centennial years) is a leap year. It is impossible to say with certainty how much is likely to go wrong, because there are so many questions about how any given computer or device will respond to date-related ambiguity. Rumors thrive on this kind of uncertainty. With the end of 1999 only months away, the predicted outcomes of Y2K-related failures range from minor inconvenience to major disaster. The reality probably lies somewhere in between.
We do know the failure of a small percentage of the world's computers could have a disproportionately large impact on financial, educational, industrial, and other systems, due to the interconnectedness of these systems. When one computer system has a Y2K failure, it can – and does – have a serious impact on the others connected to it. In test environments, this has already happened.
Laboratory and other equipment could be at risk because of embedded chips. Again, it is difficult to predict how a microprocessor will respond to date-related ambiguity. In response, MIT has hired specialists on embedded systems to assess non-computer equipment throughout the main campus. These specialists will provide information about steps that can be taken to make these devices Y2K-ready. The assessment will include not only laboratory equipment but also elevators, security systems, and other vital devices. The first phase of this process is now underway, and administrative officers will be consulted over the next few weeks about the assessment of their areas.
What about Y2K-related power failures?
Utility providers have expressed optimism about the continuity of service, yet the possibility of brownouts or blackouts has received widespread attention. Fact or hype? Unknown. However, it is known that last August's local power failure created problems in some areas of the Institute. With that in mind, precautions can be taken to preserve health and safety, as well as protect experiments and other work.
A plan for the preservation of data is an especially important precaution and is a good first step to take in preparing for Y2K. If a computer should misinterpret a date and corrupt existing data, it is vital that current backups be available to facilitate recovery. Whether a computer failure is the result of Y2K, a power failure, or other causes, the maintenance of current backups is a prudent practice.
Preservation of the data housed at MIT's Data Center is accomplished with detailed risk avoidance measures. Staffed 24 hours per day, the Center's data is backed up and taken off premises daily. Should there be a power failure, a roomful of batteries is online to keep the Data Center operations fully functional for up to 15 minutes. Within the first 5 to 10 seconds of such an outage, a 400 KW diesel generator is designed to start automatically. Should the generator fail to respond, operators have enough time to manually start the generator, or to perform a systematic shutdown of the Center's operations.
Though the Data Center is well equipped to handle data preservation, not all "mission-critical" data is stored in the Center's facilities. Data residing in departments, labs, and centers is equally important to the normal functioning of MIT and should be handled with comparable care.
Two campus resources are available to provide secure backups. The first is ADSM, an IBM software package that allows users to back up and restore, and to archive and retrieve data files on your computer over the network. An IBM mainframe VM computer is the ADSM file storage server. Information can be found at: <http://web.mit.edu/is/help/adsm>.
An off-premise data storage service is also available, and locally produced backups can be taken to a secure location. Information about this service may be obtained from the Data Center at 253-7049.
Finally, it is important to clarify some topics related to MIT's Y2K planning.
Contrary to popular belief:
* Y2K-related risks are not restricted to the domain of those who write lines of computer code. Recent polls have shown that many people believe only large computer programs are at risk, and that "someone will take care of it." In fact, large computer programs represent only a fraction of the problem. Desktop computer systems can be affected. Automobiles, alarm systems, and appliances all contain embedded chips. While it is unrealistic to expect catastrophic failures in all cases, there will be some failures.
* UPS (uninterruptible power supply) units on computers cannot keep a computer running for long periods of time. These units are useful for providing computer operators enough time to perform a systematic shutdown, but that presupposes the operator is nearby if there's a power failure. Ordinarily, over a holiday weekend like New Year 2000, it's unlikely that staff would be nearby to respond to a UPS alarm. A planned shutdown beforehand would provide better protection.
* Most alarm systems (fire, environmental, etc.) are simple notification systems. They do not respond to an outage or emergency – they simply inform those who receive the alerts that there is a problem. Here, emergency action plans are vital, and information about the names and phone numbers of those who should receive alerts must be kept up-to-date.
* Although there are more than 100 backup generators located throughout the MIT campus, with few exceptions these generators do not provide back-up power to our laboratories and offices. The generators are in place to power lights and other essentials for the safe evacuation of buildings in the event of a power failure.
Intelligent planning and sensible precautions are essential components of MIT's Y2K preparations. For advice and information, consult the Y2K Team's Website: <http://mitvma.mit.edu/mity2k>, write to y2k-help@mit.edu, or call 253-2000. Team members are also available to make meeting presentations to departments, labs, and centers.
In the months to come, this column will carry up-to-date information about Y2K concerns. Questions are welcome, and should be addressed to: y2k-help@mit.edu.
|
|
|
|
|
|
|